segmondy

35 ms·

871.
by segmondy 3y ago
For once, I really hope the US govt would do something about these sort of devices. I bought a digital picture frame from amazon. It's listed as having an SD card. When I tried to set it up. It wanted me to install an android app
872.
by segmondy 3y ago
https://hastie.su.domains/ElemStatLearn/printings/ESLII_prin...
873.
by segmondy 3y ago
this is about machine learning with Ocaml, it is very strongly related to what Janestreet does.
874.
by segmondy 3y ago
Yup, that's why I bought the 3060, my first GPU card. For the 12gb. It allows me to play with LLM. I can get 7/13/30B models, but it's dog slow. It's a terrible thing that Nvidia is dragging in increasing the
875.
by segmondy 3y ago
Studying C or Python is studying C or Python. Rather, study AI. Pick up python along the way, skip C. https://www.youtube.com/watch?v=aircAruvnKk&list=PLZHQObOWTQ... https://www.youtube.com/watch?v=jG
876.
by segmondy 3y ago
I don't think they believe that, they are just gaslighting us. Anyone that thinks so, can go to China and call our the leadership. Or go to Middle East and call out their leadership and see what happens. They can easily do that in the
877.
by segmondy 3y ago
So here's how I implemented the recovery in my proof of concept. Your backup is performed using your backup private key, and the encrypted passkeys (private keys) are backed up to the server. If you lose your device, you obviously c
878.
by segmondy 3y ago
What do you think the CEO's job is? Only to tell people what to do?
879.
by segmondy 3y ago
Why would they mention their weakness? Have you heard of Siri lately? What do they have to offer in the AI space? Nothing. At best they can contribute to ML by hardware. With that said. ML != AI. ML is a means to AI. Google, Faceb
880.
by segmondy 3y ago
Google.
881.
by segmondy 3y ago
Join the appropriate reddit groups machinelearning, locallamma, deeplearning, and follow folks in the ML space on twitter, plus use a search engine. 1. Learn to run a model, checkout llama.cpp Tons of free models on huggingface.com 2. L
882.
by segmondy 3y ago
I have never paid for a SaaS software because it's opensource. If it's opensource, I'm more likely to run it. My concern with SaaS is more around your security/data policy. Work on that, convince your customers tha
883.
by segmondy 3y ago
I use port knocking on 3 preselected ports to open up ssh which is not on port 22. I can't bother to have my logs cluttered with these garbage.
884.
by segmondy 3y ago
... strange. I'm running 30B models on a 10yr old PC with a $400 RTX 3060. Folks can run the 65B models with 4090 or dual 3090. Usually for about a cost of $2500.
885.
by segmondy 3y ago
your statement sounds like it doesn't support GPU, llama supports GPU as well. However llama is just code to run models. OP needs to find a model that works for them, there are lots of models out there.
886.
by segmondy 3y ago
Looks very neat. How do you use it? Looks like it's just to inspect a personal model and can't be applied to external models, is that right?
887.
by segmondy 3y ago
Your undo operations will be very simple. So instead of having a complex logic. Have a simple lambda function that talks to a queue. That's it. It accepts an undo command. You read a command you stuff it in the queue done. No DB
888.
by segmondy 3y ago
Weird take, design patterns take off cognitive load. It's easier to say. Use an observe or strategy pattern or adapter pattern. In most sufficiently large OO code bases, folk end up implementing design patterns sometimes and often p
889.
by segmondy 3y ago
logic is binary. yes or no. classic AI used pure logic via means of rules engine. you could tell exactly which path to take to solve a problem. fuzzy logic is not binary, but multimodal/range. what maps unto that today? probabil
890.
by segmondy 3y ago
How? Do you think they have a previous model that's missing one single data point for anything that needs to be forgotten? The only way this would work would be to filter out whatever the EU demands then retrain the model without t
891.
by segmondy 3y ago
Do you understand that NN is not like rows in database. The machine can't unlearn what it has learned. You can't make AI unlearn something. They are going to ingest large amounts of data and trying to stop it is going to be
892.
by segmondy 3y ago
It's took the form of neural networks and deep learning.
893.
by segmondy 3y ago
Yes it's possible. If you're motivated enough and willing to put the time, you can do it. There are unlimited resources on the net, find what works for you. Key is to stick to it and work through it. Get the fundamentals out o
894.
by segmondy 3y ago
intel can't even in hardware, and somehow they think they can do so with software? Right! This is just for the clueless investors.
895.
by segmondy 3y ago
Just do it. Spend a few hours doing research and you will find out. With that said, buy as much memory as can. That makes 4090 king if you have the server that can carry it, plus the budget. For me, I settled for 3060, it's a nic
896.
by segmondy 3y ago
Welp, never buying a Rivian!
897.
by segmondy 3y ago
I'm more interested in what normal folks are running at home. What are your builds?
898.
by segmondy 3y ago
Looks great! You should open this up and get code contributions.
899.
by segmondy 3y ago
I must be the only one here that loves Netflix. I probably have a 100+ TV shows and Movies in my to watch list. Anyways, $300million is a drop in the bucket compared to their yearly expense. Why is this news again?
900.
by segmondy 3y ago
I don't think they can. I believe they just add filtering in the data pipeline before & after the generation. Think about how they currently filter out "bad things" user input ---> | prompt filter | GPT | output filt

More ›